| Thu, 25 Feb 2016 18:19:27 +0000 |
Chris Cannam |
Update summary code and tests -- fix multiple outputs when requesting both summary and non-summary for the same output; and values have been changed because of the coded file cache bit depth change. Also skip network-dependent tests if there is no network.
|
| Thu, 25 Feb 2016 10:53:10 +0000 |
Chris Cannam |
Better error reporting for transform load (from RDF and XML)
|
| Fri, 20 Nov 2015 09:32:11 +0000 |
Chris Cannam |
Fix grisly error in multiplexed audio reader (calling itself!)
|
| Wed, 04 Nov 2015 10:07:29 +0000 |
Chris Cannam |
Update JAMS output to JAMS v0.2.0. We now (for the first time?!) write actual JAMS schema-compliant output when possible, though it isn't possible for many types of plugin. The output for all tested combinations of transforms is valid JSON even where it isn't schema-compliant.
|
| Tue, 03 Nov 2015 11:36:00 +0000 |
Chris Cannam |
More useful report when a transform is completely unknown to us
|
| Tue, 01 Sep 2015 17:05:32 +0100 |
Chris Cannam |
Add mandatory option --json-format to JSON feature writer, in preparation for supporting multiple JSON formats (perhaps) in future
|
| Wed, 26 Aug 2015 21:46:52 +0100 |
Chris Cannam |
Fixes from last merge
|
| Thu, 16 Oct 2014 10:32:16 +0100 |
Chris Cannam |
Change internal blocksize from 16384 to 1024 (see comment in FeatureExtractionManager ctor)
|
| Wed, 15 Oct 2014 13:02:44 +0100 |
Chris Cannam |
Add some metadata; and the data is nested more deeply than I thought
jams
|
| Mon, 13 Oct 2014 11:42:54 +0100 |
Chris Cannam |
Add -Werror in debug mode, and fix errors resulting from it
|
| Fri, 10 Oct 2014 15:33:45 +0100 |
Chris Cannam |
Remove some debug out
|
| Fri, 10 Oct 2014 15:24:28 +0100 |
Chris Cannam |
Support plugin version in RDF transforms, and bail out in Sonic Annotator if it doesn't match
|
| Fri, 10 Oct 2014 15:00:47 +0100 |
Chris Cannam |
Fail explicitly if a nonexistent plugin output is requested
|
| Wed, 08 Oct 2014 18:44:40 +0100 |
Chris Cannam |
Do not feed audio data to a plugin before its transform's start time (we might be feeding other transforms that started earlier). This breaks a test, but the test's expected output was broken (rather alarmingly)! Note that this does mean a plugin may have to-next-block "leakage" at start as well as end of its start-duration zone.
|
| Wed, 08 Oct 2014 13:14:39 +0100 |
Chris Cannam |
Tests (and minor resulting fixes) for remote file download
|
